NLPDRS: A YOLOv8 and PaddleOCR-Based End-to-End Framework for Nigerian License Plate Detection and Recognition System

Road-related crimes and traffic regulation violations remain pressing challenges in Nigeria, largely due to inadequate surveillance infrastructure and a lack of localized automatic vehicle identification systems. This study presents an end-to-end Automatic License Plate Recognition (ALPR) framework, specifically designed to address the unique structural and environmental characteristics of Nigerian license plates. A new annotated dataset of Nigerian license plates was built under diverse lighting and background conditions to support robust model training and evaluation. The proposed system leverages the YOLOv8 object detection architecture for real-time and high-accuracy license plate localization, followed by a refined Optical Character Recognition (OCR) module for alphanumeric extraction. To enhance recognition precision, image preprocessing techniques such as k-means clustering, morphological operations, and adaptive thresholding were applied before OCR inference. Experimental evaluations demonstrate that the system achieves a precision of 0.87, recall of 0.91, and a mAP@0.5 of 0.93. This study bridges the gap in domain-specific ALPR solutions by addressing Nigeria’s plate variability, non-uniform fonts, and challenging environmental conditions. The resulting system is scalable and adaptable for deployment in traffic monitoring, vehicular access control, and national security applications across Nigeria.